[NBP web-reskin] Regarding Swish indexer

[Didomenico, Steven]

Hi Ethan,

Steve DiDomenico here. There is a reason for our discussion on the Site
Search.

One direction we are currently looking at is importing text content from
a file into a page dynamically at "page request" time.

Richard Ward and I have been talking about the site search and how it
may be affected by this.

When Swish runs, how does it crawl the site? Does it just read the .html
files or does it actually render the page?

Thanks,

Steve DiDomenico
Common Impact / Fidelity Cares

-----Original Message-----
From: web-reskin-bounces at nbp.org [mailto:web-reskin-bounces at nbp.org] On
Behalf Of Ethan Rowe
Sent: Wednesday, November 25, 2009 1:51 PM
To: Discussions regarding the 2009 UI reskin of nbp.org
Subject: [NBP web-reskin] Regarding Swish indexer

All,

We've looked into the Swish index matter a bit.

The index files are built on a nightly basis in production.  They don't
exist in your camp without some preparation.  If you want to create the
index in your camp so that you won't get hard errors from the site
search, you can run:
  (for camp26, for example)
  /usr/local/bin/swish-e -c \
   /home/fidelity_camp/camp26/catalogs/nbp/etc/swish/site_search.conf \
   -S prog

That'll run for a little while and built the Swish-e index files that
the site search relies upon.

However, you're not likely to find the results all that satisfying,
because the revamped pages, from what we're seeing, mostly involve the
deletion of text content and the introduction of images.  The site
indexer focuses on text and doesn't do much with images.

Consequently, we can run the indexer, but the newly-revamped pages won't
really do great in the results because they have relatively little
content to index.  That's based on a camp built yesterday; perhaps there
are some outstanding commits that haven't been pushed upstream yet that
would change the content situation?

I think it would probably make sense to get all your reskin efforts more
settled, disregarding the search index matters for the duration, and
then let me dig into the concerns for the site search indexer at the
end.  Does that seem reasonable?

Thanks.
- Ethan
-- 
Ethan Rowe
End Point Corporation
ethan at endpoint.com
_______________________________________________
Web-reskin mailing list
Web-reskin at nbp.org
http://www.nbp.org/mailman/listinfo/web-reskin